Light Water – Run 1

17DEC00

For the first run, the cell was configured as shown in this scale drawing. The Ni wire cathode consists of 24 meters of 0.38mm dia wire (Alfa #10253) wound onto a PVDF spool that holds the cathode centered around the Pt anode (10 cm of 1 mm dia Pt wire – Alfa # 10285).

The electrolyte for the first run was 350 grams of a 0.57 molar K₂CO₃ solution made with Aldrich #24,355-8 (potassium carbonate) and distilled water. It should be noted that Mills used only 250 grams of electrolyte in his experiments but we were forced to increase this to 350 grams to get sufficient electrolyte depth in the somewhat larger Thermos Products Dewars we are using.

All interior structural components of the cell, the temperature probes, and stir bar were washed with Alconox and rinsed with distilled water. Per Mills procedure, the Dewar vessel was also rinsed with 0.1 molar HNO₃ and then again with distilled water. The cathode and anode were cleaned with the same procedures Mills used (as described in the introduction).

Throughout this run and all calibration and zero runs, the stir bar is rotated at 750 rpm by a motor-driven magnet spinning beneath the cell (not shown).

The first run was intended to emulate Mills experiment #1. The table below (reproduced from p. 479 of “The Grand Unified Theory of Classical Quantum Mechanics”, September 1996 edition), shows the pertinent data from this experiment:

Reading the first line in this table we see that experiment #1 was run with a steady (100% duty cycle) electrolysis current of 0.083 amps. In Mills’ cell that current required a voltage of 3.05 volts. This yielded a total VI input power of 0.253 watts. Mills calculated a reduced input power (V-1.48)I of 0.130 watts which would represent the net power deposited in the cell if all the H₂ and O₂ produced by the electrolysis escaped from the cell in elemental form (i.e. did not recombine into H₂O inside the cell).

Mills observed a thermal output power of 0.275 watts with his calorimetry. He shows that this is 0.145 watts higher than the reduced input power and he shows that the ratio of output power to the reduced input power is 2.12, or 212%.

It must be emphasized that these last figures (the excess power and the output/input ratio) are based upon the assumption that no recombination of the electrolysis gas was taking place inside the cell…i.e. that all of its caloric value was escaping. Mills based this assumption on a separate cell that was specifically designed to allow the electrolysis gases to be captured and measured (see p 474 of the book referenced above). Apparently Mills’ Dewar cells, in which the active experiments were conducted, did not permit such gas measurements.

Results of our Experiment

This plot depicts the results of Run 1. The horizontal scale is time and covers 300 hours (10 hr/div). Six different parameters are plotted vs time: Room temperature (white jagged line near the bottom), reference cell temp (wavy blue near bottom), and active cell temp (yellow) are all plotted on a vertical scale that runs from 20 to 40 °C (2°C/div).

Cell current (dim gray line near the top) is plotted on a scale from 0-100mA and cell voltage (blue line middle) is plotted on a scale from 0-5 volts).

The most important lines are Pin (green) and Pout (purple). Pin (input power) was calculated as Mills did using (V-1.48)I. Pout (thermal output power) is a linear function of the temperature difference between the active and reference cells, i.e. Pout = (dT + offset)/R. The “offset” is a small adjustment used to compensate for the difference between the temperature sensors in each cell (in this case, offset = 0.0295°C) and “R” is the thermal resistance of the Dewar+lid enclosure. The value of R for this plot was 36.23 °C/watt.

For the first 80 hours of this run, only electrolysis power was applied to the cell. Pin was about 114 mW during this time and Pout gradually rose and leveled off at about 191 mW, indicating an apparent excess power of 77 mW (see discussion of possible artifacts below). At about 80 hours (8 horiz. divisions into the run), we began dissipating 100.5 mW of power in the calibration resistor immersed in the electrolyte of the active cell. After another 80 hours had elapsed, the Pout signal had risen to about 291 mW, 100 mW higher than before the resistor was energized. At 170 hours we turned off the calibration resistor and observed that the system relaxed very close to its original state.

It should be noted that the R value used for this data (36.23 °C/watt) was empirically adjusted to make the standard addition of 100 mW report correctly in the Pout signal. It is about 10% lower than the R value we determined earlier using pure water (and no electrolysis) in the Dewar. We do not know the cause for this shift but surmise that it has to do with the fact that electrolysis generates a fine mist which carries electrolyte to the upper surfaces of the Dewar vessel which are not normally wetted. The mist would warm these surfaces continuously and thus increase the average delta-T across the Dewar walls. This hypothesis remains to be confirmed.

Comparison of Results

This table shows our results for Run 1 side-by-side with Mills’ results for his Experiment #1. The most important difference between our results is the output power. Mills got 275 mW, which is larger even than his VI input power of 253 mW. We got only 191 mW, which is less than our VI input power of 236 mW. We have no explanation for this significant discrepancy.

However, using the (V-1.48)I input power that Mills used, our results still show 77 mW of excess power for an output/input ratio of 168%. But remember that the (V-1.48)I input power is only correct if all of the electrolysis gases produced in the cell actually escape from the cell.

Unfortunately, we do not know what fraction of the electrolysis gas was recombining in the cell during Run 1.

Gas Flow Measurements

During the ~ 300 hour duration of Run 1, we measured the volume flow rate of the electrolysis gas escaping from our cell many times using the apparatus pictured. At face value, these measurements indicated a very high degree of recombination (i.e. ~ 90%) occurring within the cell. However, recent developments have shown that these measurements are certainly in error.

Recent measurements (on a new run) indicate that the degree of recombination occurring in the cell is approximately 50% but that is a very rough estimate and we are still far from understanding what appears to be an impossible situation with our gas flow measurements.

The gas flow measurements were made by immersing a glass tube vertically into a container of liquid as shown in this photo. The liquid rises up inside the glass tube matching the level outside. When the gas is admitted to the upper end of the glass tube, the meniscus between gas and liquid inside the tube begins to fall, pushed down by the incoming gas. We simply use a stopwatch and a scale next to the vessel to measure the time required for the meniscus to fall 1 centimeter. The inside diameter of this 4 mm glass tubing is 2.23 mm. A 1 cm length of this tubing therefore has a volume of 0.039 cm³.

In the early stages of Run 1, we studied the dependence of the gas flow measurement on backpressure (and explored the possibility that a small leak was present in the system). We arranged a long glass tube immersed vertically in 30 cm deep water (in a tall graduated cylinder) and measured the 1cm drop time at various points throughout the 30 cm depth. In other words, we measured the exiting gas flow rate while varying the back pressure from 0 to 30 cm of water. There was no discernible reduction in flow rate with increasing head pressure. The expected 3% reduction (from ideal gas consideration of the increasing pressure) was lost among the ~10% temporal variations. More importantly, the flow rate showed no signs of stopping even at 30 cm of head pressure, which it surely would have had there been a small leak in the system.

Now here’s the “impossible” part: When we switched flowmeter liquids from water to Hg, the observed flowrate decreased dramatically (i.e. by a factor of ~10), even when the Hg meniscus was <1 cm below the surface of the Hg in the vessel! This seems impossible because the earlier tests with 30 cm deep water proved that the gas flow rate was more-or-less independent of back pressure up to 30 cm of water. Since Hg is 13.6 times denser than water, we should therefore expect similar behavior up to a depth of 2.2 cm in Hg…but no! Something causes the gas flow to “instantly” reduce by ~10 (suspiciously close to the 13.6 density ratio) when the liquid is switched from water to Hg. We also tried using methanol as a flowmeter liquid and, sure enough, the observed flow rates with methanol (s.g.= 0.8) are about 20% higher than with water.

Most recently (on a new run), we have started making measurements with a “headless” measurement system that consists of a horizontal section of the same 4 mm glass tubing that is sealed by a small “plug” of water. When the electrolysis gases are fed into the end of this tube, the plug is forced along horizontally like a liquid piston. If the walls of the tube are wetted, the plug moves essentially effortlessly. These measurements indicate a gas flow rate about 5 times higher (i.e. indicating 50% recombination in the cell) than those obtained from the vertical tube apparatus with water as a working fluid.

Calibration Details

This plot shows the results of the calibration we performed with the active cell filled with 350 grams of pure H₂O and power dissipated in the cell via the calibration resistor. The two calibration points bracket the power range that the actual experiment employs and they indicate an adequate degree of linearity in the calorimeter response.

Furthermore, the method of standard addition employed during Run 1 provides a confirmation that the calorimeter is responding accurately to known power levels.

The calorimeter offset was determined on a special zero run (conducted before Run 1) using pure H2O in the cell. Adhering strictly to Mills procedure, which specifies that the Ni cathode never be left in the electrolyte without electrolysis current flowing, we were unable to check the zero reading with the cell assembled as it was for Run 1. This is not expected to cause a noticeable error.

It is conceivable that thermal heterogeneity inside the cell could cause the response to electrolysis heat to be significantly different than the response to heat from the calibration resistor but this can be ruled out with a detailed examination of the readings from the three electrolyte temperature probes in the active cell.

This plot of the data from Run 1 shows how the readings from the three electrolyte temperature sensors varied with respect to each other. Each of the three thick colored lines represents the difference between one sensor and the average of the three sensors. Zero is in the center of the graph and the top and bottom represent +0.1°C and -0.1°C respectively. The Pout curve (purple) is included for reference so the time during which the standard addition was made can be easily identified.

First we see that all three sensors are individually within 0.1°C of the mean. This is not surprising since these particular thermistors are guaranteed by BetaTHERM to be accurate within 0.2°C and their readings are presented at face value (i.e. without any adjustments).

More importantly we see that there is little or no perturbation in the readings of these sensors when the 100 mW standard addition is turned on. These data demonstrate that there is no significant thermal heterogeneity within the electrolyte.

Post-run Examination

After the run was completed we disassembled the cell and examined the electrodes and electrolyte. Surprisingly, their appearance was essentially unchanged by the nearly 300 hours of electrolysis. The electrolyte was perfectly clear with no sign of precipitate on the bottom of the vessel. The Ni cathode looked much the same as it did at the start of the run and so did the Pt anode. We also performed a qualitative XRF analysis of the used electrolyte and observed no signs of Pt or Ni dissolved in it (we would have noticed as little as 10 ppm of these elements).

Conclusion

Perhaps the most significant result of this experiment is the difference between Mills observed output power of 275 mW (which is larger than his VI input power of 253 mW) and our output power of 191 mW (which is less than our VI input power of 236 mW). We have no explanation for this significant discrepancy.

Lacking accurate gas flow measurements on Run 1, we simply cannot make a quantitative evaluation of the power balance on this run. However, as an exercise let us assume that our observed output power of 191 mW is correct and that the cell is not producing excess power. With those assumptions we can calculate what fraction of the electrolysis gas should be escaping the cell from the relation (V-1.48x)I = 191 mW. This yields x = 0.37 or that 37% of the electrolysis gas should be escaping from the cell. In view of our recent gas flow measurements, we can only say is that this value may have been the actual flow rate during Run 1.

We will endeavor in future runs (underway now) to obtain accurate gas flow measurements using the “headless” gas flow apparatus. Meanwhile, we would greatly appreciate assistance in explaining the “impossible” behavior (described above) exhibited by our original gas flow apparatus.